No. 2, 2020
AI for Good: Global impact Editorial ITU News MAGAZINE 02/2020 1
AI for Good Global Summit: Now stronger than ever
Houlin Zhao
ITU Secretary‑General
J Artificial Intelligence (AI) has ITU continues to draw upon enormous potential to accelerate expertise from the AI for Good progress on all 17 of the United Global Summit community Nations Sustainable Development and has launched an AI for Goals. Good Webinar Series delving into promising use cases of AI in Every year, we see progress healthcare, smart mobility, food across the world. But so much supply, gender inclusivity and more can be done. environmental global chal‑ lenges, including how to com‑ This continuous That’s why ITU joins forces with bat COVID‑19. approach is partners each year to put on the allowing us to AI for Good Global Summit. In this edition of ITU News broaden the Magazine, you will find a range Now in its fourth year, the Summit of useful expert insights on how AI for Good has gone virtual due to COVID‑19 AI is being leveraged responsibly movement travel restrictions. It is now all year, to improve lives — and how ITU is wider than always online. weaving relevant AI for Good ini‑ tiatives into its work to ensure our ever before. This continuous approach is Union is doing all it can to seize allowing us to broaden the AI the opportunities at hand. Houlin Zhao for Good movement wider than ever before. Contents ITU News MAGAZINE 02/2020 2
AI for Good: Global impact
Editorial
1 AI for Good Global Summit: Now stronger than ever Houlin Zhao ITU Secretary‑General Cover photo: Shutterstock
AI for Good Global Summit
4 Follow us at the AI for Good Global Summit
5 2020 AI for Good Global Summit to scale AI‑powered problem solving for global impact
9 International standards for an AI‑enabled future ISSN 1020–4148 ITU News caught up with Chaesub Lee, Director of the ITU itunews.itu.int Telecommunication Standardization Bureau, to learn more about Six issues per year the latest ITU standards projects addressing AI and Machine Copyright: © ITU 2020 Learning and the value of the AI for Good Global Summit. Editor-in-Chief: Matthew Clark 13 AI for Good — Time to move the needle Art Editor: Christine Vanoli Editorial Assistant: Angela Smith By Fred Werner Head of Strategic Engagement, ITU Standardization Bureau Editorial office: Tel.: +41 22 730 5723/5683 E‑mail: [email protected] 18 Join the global initiative on ‘AI and Data Commons’ Mailing address: International Telecommunication Union Place des Nations AI for Good goes digital CH–1211 Geneva 20 (Switzerland)
20 COVID‑19: How Korea used innovative technology and Disclaimer: AI to flatten the curve Opinions expressed in this publication are those of the authors and do not engage ITU. The des- By ITU News ignations employed and presentation of mate- rial in this publication, including maps, do not 24 COVID‑19: Where are the self-driving cars imply the expression of any opinion whatsoever and trucks? on the part of ITU concerning the legal status of any country, territory, city or area, or concerning By ITU News the delimitations of its frontiers or boundaries. The mention of specific companies or of certain 27 Tech startups join AI for Good Innovation Factory with live products does not imply that they are endorsed pitching or recommended by ITU in preference to others of a similar nature that are not mentioned. By ITU News All photos are by ITU unless specified otherwise. Contents ITU News MAGAZINE 02/2020 3
AI for sustainable development
31 Africa’s AI community set to light up the AI for Good Global Summit By Alexander Tsado and Nick Bradshaw Alexander Tsado, Advisor for Alliance4AI and Nick Bradshaw, Co-Founder of AI Expo Africa
34 How can we solve the problems of gender bias in AI? Experts weigh in. By ITU News
37 COVID‑19: How can AI help fight hunger and food insecurity? By ITU News
42 How AI can help save the planet By ITU News
46 What the United Nations is doing to leverage AI for Good By ITU Secretary-General Houlin Zhao
AI and health care
50 Reducing the global healthcare shortfall using AI By ITU News
54 The opportunities and challenges of introducing AI to improve health care By Chaesub Lee Director of the ITU Telecommunication Standardization Bureau
57 Responsibly unlocking the potential of data in health By Stefan Germann and Ulla Jasper Stefan Germann, CEO and Ulla Jasper, Policy Officer, Fondation Botnar AI for Good Global Summit ITU News MAGAZINE 02/2020 4
Follow us at the AI for Good Global Summit AI for Good Global Summit ITU News MAGAZINE 02/2020 5 Shutterstock
2020 AI for Good Global Summit to scale AI‑powered problem solving for global impact
J Ten years remain to achieve the United Nations Sustainable Development Goals (SDGs). Leaders in AI and humanitarian action will convene at the 2020 AI for Good Global Summit with strong intent to Three editions of the AI ensure that “AI for Good” solutions achieve a scale matching that of the for Good Global Summit ambitions captured by the SDGs. have recognized the significance of the The AI for Good Global Summit Now in its fourth edition, this leap from AI promise is the leading United Nations year’s AI for Good Global Summit to global impact. platform for inclusive dialogue on is being held online all year, AI. The summit identifies practical and will continue to connect applications of AI to accelerate AI innovators with public and Houlin Zhao progress towards the SDGs and private-sector decision-makers ITU Secretary-General builds collaboration to assist in the interests of stimulating the these applications in achieving discovery and delivery of “AI for global impact. Good” solutions for all.
Join the conversation on social media using the hashtag #AIforGood AI for Good Global Summit ITU News MAGAZINE 02/2020 6
The 2020 summit is co-or‑ “Three editions of the AI for Good Speakers ganized by the International Global Summit have recognized Telecommunication Union (ITU) the significance of the leap from Over 100 speakers are expected and early confirmations include: — the United Nations special‑ AI promise to global impact,” ized agency for information said ITU Secretary-General ` Yoshua Bengio — Founder and Scientific and communication technol‑ Houlin Zhao. “We see renewed Director, Montreal Institute for Learning Algorithms (MILA); ACM Turing Award ogies (ICTs) — and the XPRIZE resolve within the AI for Good Winner 2018 Foundation, in partnership with community to create the condi‑ Switzerland, the Association for tions necessary to make this leap ` Anousheh Ansari — CEO, XPRIZE Foundation; Space Ambassador Computing Machinery (ACM) and accelerate progress towards and a wide variety of sister United the achievement of the SDGs.” ` Stuart Russell — Professor, UC-Berkeley; Nations agencies. Adjunct Professor, UC-San Francisco; Author of “Human Compatible AI” “With only 10 years remaining to The 2017 summit marked the achieve the United Nations SDGs, ` Stella Ndabeni-Abrahams — beginning of a global dialogue AI is sure to be an invaluable tech‑ Minister of Communications and Telecommunications, South Africa on the potential of AI to act as a nology to help accomplish these force for good. The action-ori‑ critical goals that are designed ` Francesca Rossi — IBM AI Ethics Global ented 2018 and 2019 summits to ensure a more sustainable Leader gave rise to numerous “AI for and equitable future for all,” said ` Pascale Fung — Professor, Department Good” projects including an XPRIZE CEO Anousheh Ansari. of Electronic & Computer Engineering, “AI for Health” Focus Group led “At XPRIZE, our mission is to Hong Kong University of Science & by ITU and the World Health facilitate radical breakthroughs Technology; Director of Centre for Artificial Intelligence Research (CAiRE) Organization, an ITU Focus Group for the benefit of humanity, on “AI for Autonomous and and our ongoing collabora‑ ` Shwetak Patel — Washington Research Assisted Driving”, and an open tion on the AI for Good Global Foundation Entrepreneurship Endowed Professor in Computer Science and framework for collaboration in Summit is one way we hope Engineering and Electrical Engineering “AI Commons”. to accelerate safe and ethical at the University of Washington; ACM development of AI and machine Computing Prize Winner 2018 learning technology.” ` Maggie Carter — Principal, Disaster The pursuit of Response Program, Amazon Web global impact The summit attracts a unique Services (AWS) cross-section of AI experts from ` Elizabeth Bramson-Bodreau — CEO, MIT The pursuit of global impact will industry and academia, global Technology Review be the defining feature of the business leaders, Heads of United ` Vicki Hanson — CEO, ACM 2020 summit. Nations agencies, ICT ministers, non-governmental organizations, ` Anja Kaspersen — Director, United civil society and artists. Nations Office for Disarmament Affairs ` Lucas di Grassi — Formula E Racing Champion; CEO, Roborace; United Nations Clean Air Ambassador
Please visit the summit website for more information on the latest speakers, breakthrough sessions, solution tracks and demos. AI for Good Global Summit ITU News MAGAZINE 02/2020 7
Generating “AI for Good” Cutting edge R&D projects and breakthroughs Cutting edge AI research The summit is designed to and development and AI inspired generate “AI for Good” projects performances from ground able to be enacted in the near breaking artists and musicians term, guided by the summit’s will be featured as virtual live multi-stakeholder and inter-dis‑ demos now that the Summit has ciplinary audience. It also aims to gone digital. ensure trusted, safe and inclusive development of AI technolo‑ gies and equitable access to their benefits. Sponsors The 2020 summit aims to gener‑ ate AI breakthroughs in climate and environmental action, the elimination of hunger, gender Organization and equality, health care, smart and partnership safe mobility, the preservation of cultural heritage and the pro‑ tection of access to trustworthy The 2020 summit is kindly information. supported by sponsors.
An “AI Innovation Factory” will Strategic Sponsor showcase new “AI for Good” Switzerland ideas, crowd-sourced AI chal‑ The summit is organized Gold sponsors lenges and promising AI start‑ in partnership with the ACM ups. A “partner day” will offer following sister United Kay Family Foundation potential AI adopters an audi‑ Nations agencies: Fondation Botnar ence with leading AI experts ETRI CTBTO; FAO; ICAO; Immersion 4 and educators. ILO; IMO; IOM; UNAIDS; UNCTAD; UNDESA; Supporters UNDPA; UNECE; UNEP; Microsoft UNESCO; UNFCCC; TRAX.GD UNFPA; UNGP; UN Pink Lion Habitat; UNHCR; UNICEF; Crowd 4 SDG UNICRI; UNIDIR; UNIDO; Monash University — UNISDR; UNITAR; UNODA; Monash Data Futures UNODC; UNOOSA; Institute UNRISD; UNU; UN Women; UNWTO; WBG; WFP; WHO; WIPO; and WMO. AI for Good Global Summit ITU News MAGAZINE 02/2020 8
3 ’Breakthrough’ Tracks and 4 ’Solutions’ Tracks The AI for Good Global Summit contains “Breakthrough” sessions and “Solutions” tracks designed to provide focus in key areas.
AI for Gender AI for Food AI for the Equality The AI for food Environment The AI for breakthrough The AI for the gender equality session tackles environment breakthrough session issues around food breakthrough session will addresses current gender insecurity, including explore how AI solutions can issues in AI, including but not but not limited to the impact of shed light on environmentally destructive limited to algorithmic bias and data COVID‑19 on food systems. The goal practices as well as protect our world. The inclusivity. The goal of this session is of this session is to identify how goal of this session is to take urgent action to identify AI solutions that empower AI can help us reach zero hunger in combatting climate change, to identify underrepresented communities and shape the food landscape of solutions to sustainably use our planet’s and enable an equitable future for tomorrow. resources, and to re-imagine the future of humanity. the Earth with AI. Breakthrough
AI for Trust Solutions AI for the Preservation of Culture The AI for trust solutions track and Natural Heritage will establish a concrete, inclusive The AI for the preservation of culture and natural and actionable discussion on heritage solutions track intends to facilitate the how AI‑enabled solutions can help matchmaking of these ideas, challenges and AI achieve the Sustainable Development solutions among the participating stakeholders. This track Goals for peaceful, well-informed sustainable will accommodate AI and data science experts and academics, communities, the advancement of public access cultural heritage institutions coming from various countries, and to truthful information, and the protection of experts with relevant practical use cases. The overarching goal fundamental freedoms as related to the digital is to advance the area of AI for culture and historical heritage society. preservation.
AI for the Future of AI for Health Solutions Smart Mobility AI can improve the quality of affordable services, The AI for the future of smart optimize the distribution of resources in mobility solutions track will underdeveloped and understaffed communities establish a concrete, inclusive and and create inclusive and responsive solutions for actionable discussion on how AI‑enabled health care, diagnosis, triage or treatment decisions. solutions can help achieve the Sustainable However, it is not enough to deploy these solutions in the wild. Development Goals for improving in-transit traffic We must also examine how these applications can elevate the quality safety, developing energy-/environment-efficient of work and life that medical practitioners have in the field, including vehicles, transportation systems and infrastructure but not limited to reducing burnout and creating more avenues and providing inclusive and accessible mobility for better work-life balance. To this end, The AI for health solutions opportunities for all. track will extend this work, by identifying new ways in which AI can improve healthcare services and the profession at large. Solutions AI for Good Global Summit ITU News MAGAZINE 02/2020 9
International standards for an AI‑enabled future
ITU News caught up with Chaesub Lee, Director of the ITU Telecommunication Standardization Bureau, to learn more about the latest ITU standards projects addressing AI and Machine Learning and the value of the AI for Good Global Summit.
Where do we stand today in The scope of debate around AI terms of AI applications and how extends far beyond the scope of might that evolve? any single organization. This is why ITU has called for an inclu‑ ^ Innovation by and of AI is sive global dialogue on the Alongside recent accelerating, and this is evi‑ implications of AI for the future breakthroughs, denced by the contributions driv‑ of our society — a dialogue that new partnerships ing ITU’s work. AI and Machine is anchored by the AI for Good are also Learning are gaining a larger Global Summit. share of the ITU standardization supporting work programme in fields such What value do ITU and the growing as network orchestration and broader “AI for Good” community confidence management, multimedia coding, draw from the AI for Good Global service quality assessment, opera‑ Summit? in AI. tional aspects of service provision and telecom management, cable ^ Alongside recent break‑ Chaesub Lee networks, digital health, environ‑ throughs, new partnerships are Director, ITU mental efficiency, and autono‑ also supporting growing con‑ Telecommunication mous driving. fidence in AI. The AI for Good Standardization Bureau Global Summit continues to offer But AI and Machine Learning are valuable support to the “AI for finding very practical applications Good community” in creating and across industry sectors — applica‑ sustaining these partnerships. tions with considerable potential to act as a force for good. AI for Good Global Summit ITU News MAGAZINE 02/2020 10
Experts from different fields Could you share more insight into are coming together to align the aims of these initiatives? incentives for innovation and solve problems with AI. We see ^ The ITU Telecommunication connections forming among AI Standardization Sector (ITU–T) specialists, AI users, data owners Study Groups are where ITU and experts in various domains members work together to The ITU Focus Group on to benefit from AI applications develop international standards. “Environmental Efficiency — domains where AI could make for AI and other Emerging key contributions to sustainable ITU–T Focus Groups are flexible Technologies” will benchmark development. structures that are operational for best practices and describe a short period of time (typically pathways towards a standardized The United Nations Sustainable 1–2 years). They accelerate stud‑ framework to assess environmen‑ Development Goals (SDGs) ies in fields of growing strategic tal aspects of the adoption of provide the guiding light to this relevance to the ITU membership. emerging technologies. innovation. Open to all interested parties, these groups prepare a basis for Inclusive dialogue helps all stake‑ related standardization work in holders to build an understand‑ ITU–T Study Groups. ing of their respective roles in nurturing ICT innovation. This dia‑ Let me highlight five open plat‑ logue supports the development forms advancing various aspects of new partnerships and clarifies of AI and Machine Learning. The ITU Focus Group on “AI for the contributions expected of Health”, driven in close collabora‑ various stakeholders, including tion by ITU and WHO, is working the contribution expected of ITU towards the establishment of standardization. For example, a framework and associated the motivations behind initiatives processes for the performance such as the ITU Focus Groups on benchmarking of “AI for Health” “AI for Health” and “AI for auton‑ solutions. omous and assisted driving” and The ITU Focus Group on “Machine the new Global Initiative on “AI Learning for Future Networks and Data Commons” were first including 5G” is defining the elaborated at the AI for Good requirements of machine learning Global Summit. as they relate to interfaces, proto‑ cols, algorithms, data formats and network architectures. AI for Good Global Summit ITU News MAGAZINE 02/2020 11
How are ITU standards These “Machine Learning for addressing AI and Machine 5G” standards are also guiding Learning? contributions to a new ITU Global Challenge on AI and Machine ^ Information and communica‑ Learning in 5G. tion technology (ICT) companies The ITU Focus Group on “AI in the networking business are AI and Machine Learning play for Autonomous and Assisted introducing AI and Machine an important role in multimedia Driving” is working towards the Learning as part of their innova‑ coding, an area of ITU standards establishment of international tions to optimize network oper‑ work known for the Primetime standards to monitor and assess ations and increase energy and Emmy winning video-com‑ the behavioural performance of cost efficiency. pression standards, ITU H.264 the AI “drivers” in control of auto‑ MPEG‑4 Advanced Video Coding mated vehicles. New ITU standards provide an (AVC) and High Efficiency Video architectural framework for the Coding (HEVC, published as ITU integration of Machine Learning H.265 | ISO/IEC 23008‑2), as well into 5G and future networks (ITU the upcoming Versatile Video Y.3172), a framework to evaluate Coding (VVC) to be complete intelligence levels across dif‑ before the close of 2020. ITU has ferent parts of the network (ITU also established a new working Y.3173), and a framework for data group (“Question”) on “Artificial The new Global Initiative on handling in support of machine intelligence-enabled multimedia “AI and Data Common”, estab‑ learning (ITU Y.3174). applications” (Q5/16). lished in January 2020, aims to support AI for Good projects in achieving global scale. The Initiative will offer assemblies of resources to launch new AI pro‑ jects aligned with the SDGs, and scale them up fast. AI for Good Global Summit ITU News MAGAZINE 02/2020 12 Shutterstock
AI and Machine Learning are The lessons learnt developing A new ITU standard provides the widely used in developing ITU P.1203, P.1204, E.475 and framework for a premium cable models to assess the quality of P.565 will be presented by an network platform to support speech, audio and video, for upcoming ITU Technical Report industry in offering advanced example in ITU standards for the and Supplement. multimedia services (ITU J.1600). quality assessment of audiovis‑ It is the first of a new series of ITU ual streaming, in particular ITU Other notable new ITU stand‑ standards on AI‑assisted cable P.1203 (progressive-download ards relevant to AI and Machine networks. and adaptive-bitrate AV) and ITU Learning address environmental P.1204 (video streaming services sustainability, cable networks, AI is one of the five character‑ up to 4K). and operational aspects of istics of a new ITU framework service provision and tele‑ to support smart service oper‑ New ITU standards address com management. ation, network management intelligent network analytics and and infrastructure maintenance diagnostics (ITU E.475) and the A new ITU standard specifies a (ITU M.3041). New ITU stand‑ creation and performance testing datacentre infrastructure manage‑ ards under development in this of Machine Learning-based mod‑ ment (DCIM) system based on domain will address AI‑enhanced els to assess the impact of the Big Data and Artificial Intelligence telecom operation and man‑ transmission network on speech technology” (ITU L.1305), sup‑ agement, energy saving for 5G quality for 4G voice services (ITU porting DCIM systems in reducing Radio Access Networks with AI, P.565). the energy required to control and robot-based smart patrols of datacentre temperature. telecoms networks. AI for Good Global Summit ITU News MAGAZINE 02/2020 13 Shutterstock
AI for Good — Time to move the needle
By Fred Werner
Head of Strategic Engagement, ITU Standardization Bureau
J Artificial Intelligence (AI) has So what is good? great potential to help us solve We must be humanity’s biggest challenges. Different societies have different vigilant that AI From combatting climate change priorities, and a different under‑ develops in a safe, to cleaner energy to affordable standing of what is “good”. So secure, trusted health care and global pandemic how do we know what global response, the potential is there. challenges to work on? That and inclusive However, our race to capture is easy... We have the United manner for all. value from the technology Nations Sustainable Development challenges our ability to fully Goals (SDGs) to guide us. A set leverage AI to improve our of goals to improve the quality Fred Werner quality of life and the world we and sustainability of life on Earth Head of Strategic Engagement, live in. In order to use AI to make by 2030. ITU Standardization Bureau a difference, we must use AI for Good. AI for Good Global Summit ITU News MAGAZINE 02/2020 14
For example, AI can help: At a minimum, we are looking at a You can 10-year timeline, bringing us right argue so far that our 1.7 billion unbanked up to 2030. main solution for AI is individuals gain access to to understand consumer digital financial services; We have to act now if we want to preferences better so we move the needle. can influence their tastes reduce 1.3 million deaths and opinions. We do that annually on our roads; in ways where companies Scaling AI for Good create monopolies around translate and educate in 2000 the data collected on your tastes. We concentrate the African languages; There is no shortage of inno‑ value coming from those vative AI for Good applications applications to very few monitor and and use cases; from the use of hands. That, ladies and protect ecosystems; smartphones for early diagnosis gentlemen, cannot be the of disease and pandemic contact smartest way of using lower public health costs tracing, to robotics for increased Artificial Intelligence. for millions; agricultural productivity, machine learning for enhanced cybersecu‑ elevate the quality of data rity and optimized telecommuni‑ Jim Hagemann Snabe collection during pandemics cation networks. Chairman of the without sacrificing privacy; and Board, Siemens improve the quality and accessibility of civic services in overcrowded cities.
Moving the needle We need The potential is undoubtedly to ignite a bigger there, but we are running out of conversation centered time. We have 10 years left to around how to use AI achieve the 17 SDGs. AI solutions and data to help find scalable solutions to that we identify today need a the biggest challenges few years to develop, a few more facing humanity. years to achieve scale, and then a few years after that to achieve the desired impact. Anousheh Ansari CEO, XPRIZE Foundation AI for Good Global Summit ITU News MAGAZINE 02/2020 15
However, it is one thing to problem-solving power of half the develop a solution in a high-tech planet. It is crucial that we connect Project lab and another thing to deploy the remaining 50 per cent, so that Zero’ will work towards and scale these solutions across we can start firing on all cylinders. the establishment developing countries, being of a ’global service mindful of harsh conditions on platform’ — an enabling the ground and the societal, The eye in the sky infrastructure and financial and political chal‑ common capabilities — built on the automated lenges involved. A number of the SDG targets analysis of satellite could help to be achieved by imagery, opening access Connecting “problem owners” tracking from space. For exam‑ to the infrastructure with “AI innovators” needs to be ple, AI-powered satellite imagery required for associated as easy as ordering an Uber if we analysis can be used to predict AI services to achieve are serious about scaling AI for and prevent deforestation, track immediate global reach. good. We need to help people livestock with great accuracy, map speak the same language and poverty, provide data analytics identify open algorithms and for micro-insurance to small-hold Stuart Russell publicly available data sets to farmers. Professor of Computer help them solve their challenges. Science, University of This is a potential gamechanger California, and Adjunct Professor of Neurological The world needs an AI and Data that requires massive-scale collab‑ Surgery, University of commons as an enabling plat‑ oration and significant funding. California, Berkeley form to scale AI for Good prob‑ If we cannot see the needle, we lem solving. cannot move it.
Firing on all cylinders Will we become irrelevant?
We have reached a landmark AI is an extremely powerful tech‑ where half the world’s population nology that is not without its own is online. While some might see risks and challenges. We must this as an amazing achievement, be vigilant that AI develops in a the fact remains that 50 per safe, secure, trusted and inclu‑ cent of the population remains sive manner for all. We must be unconnected. This is the equiv‑ mindful of inherent biases already alent of a V8 engine only firing baked into our systems and avoid on four cylinders. We are not unintentionally codifying the benefiting from the shared art, worst of human behaviour into culture, music, creativity, knowl‑ future algorithms. edge, wisdom and potential AI for Good Global Summit ITU News MAGAZINE 02/2020 16
Will AI put us all out of work or The goal of the Summit is to iden‑ even worse, make us irrelevant? By bringing tify practical applications of AI to AI experts themselves say that AI together AI technologists achieve the SDGs and scale those is too important to leave it to the with leaders in solutions for global impact. experts alone. This issue affects government, industry, every person, every company, and humanitarian The Summit has delivered on its every institution, and every gov‑ initiatives, new ways action-oriented promise, giving ernment. It is imperative that we to apply AI to pressing rise to the AI Commons and world challenges are bring as many voices as possible generating numerous AI for Good imagined and realized. to the table. projects in fields including edu‑ cation, health care and wellbeing, social and economic equality, Vicki Hanson What do we want? space research, and smart and CEO, Association for safe mobility. Computing Machinery (ACM) Through all of this, we should not lose sight of what is humanity, Additionally, the Summit has our own intelligence and what it generated the new ITU Focus is we truly want. It is often easier Group on “AI for autonomous and to blame technology, focusing on assisted driving” that will work our fears and “what if” scenarios towards the establishment of rather than discussing our core We need to find innovative ways international standards to monitor values and charting a beneficial to incentivize and mobilize the and assess the performance of path forward for humankind. If power of the crowd, combined the AI “Drivers” steering auto‑ we do not know what we want with AI to unlock new break‑ mated vehicles. for our future, how can we move throughs and solutions. the needle? Work continues on projects that were ideated at earlier summits, Pathways forward such as the ITU Focus Group on Audacious challenges Artificial Intelligence for Health So where do we begin? The AI (FG AI4H) with WHO, working Many current global challenges for Good Global Summit is the towards the establishment of seem impossible to solve and leading action-oriented, global & a framework and associated companies, institutions and gov‑ inclusive United Nations platform processes for the perfor‑ ernments alike do not have the on AI. The Summit is organized by mance benchmarking of “AI for means or the will to tackle them the ITU with XPRIZE Foundation, Health” algorithms. head on. Solutions can actually in partnership with 36 UN sister come from anyone, anywhere. agencies, ACM and our strategic partner Switzerland. AI for Good Global Summit ITU News MAGAZINE 02/2020 17
All year, always online The digital edition of the AI for Connect — Ensure that the Good Global Summit has already remaining 50 per cent of the Due to recent developments begun with the launch of the AI world can fire on all cylinders. concerning COVID‑19, the 2020 for Good Webinar series, AI for edition of the AI for Good Global Good Innovation Factory, weekly Be vigilant — Look out for Summit will now be presented as AI for Good artists, and more. As inherent biases, safety and a continuous digital event, featur‑ the year progresses, the Summit security risks. ing weekly programming across will make its way through the multiple formats, platforms and many confirmed AI for Good ses‑ Monitor — Use real-time time-zones, including keynotes, sions and speakers from the 2020 tracking to monitor our expert webinars, project pitches, Summit programming, and will progress towards achieving Q&As, performances, demos, also tackle more region-specific the SDGs. interviews, networking and more. content. Humanize — Focus on our own We see this as an opportunity to intelligence and what we really scale AI for Good and reach even The time is now to… want for our future. more people, supporting our goal of being the most diverse Act — Create practical AI Move the needle — Employ and inclusive platform around for Good solutions aligned innovative problem solving beneficial AI. With a wider and with the SDGs through the methods to bring about more inclusive outreach, as well breakthrough sessions and radical breakthroughs for the as year-long visibility, our new innovation factory. benefit of humanity. event format provides partners, speakers and supporters with Scale — Use the Global a much larger, more visible Initiative on AI and Data opportunity to connect problem Commons as an enabling owners with AI problem solvers platform to scale AI for Good. and work together on actionable projects that shape the future of AI for Good. AI for Good Global Summit ITU News MAGAZINE 02/2020 18 Shutterstock
Join the global initiative on ‘AI and Data Commons’
J Datasets relevant to The Global Initiative is supporting Your responses will assist the ini‑ sustainable development, the AI for Good community in tiative in matching your interests expertise in AI and humanitarian overcoming these challenges. to projects, working groups and action, and “AI for Good” projects new e‑meetings. with ambitions to achieve global Impact on a global scale will impact — these are the three key require common enabling pillars of the Global Initiative on infrastructure — “AI and Data An open framework AI and Data Commons. Commons” — comprising shared for collaboration knowledge, datasets and The AI for Good Global Summit problem-solving approaches The Global Initiative is an open has showcased numerous AI to stimulate the develop‑ framework for collaboration projects with promise to advance ment and application of AI for akin to the Commons it seeks to the United Nations Sustainable Good projects. create — a decentralized system Development Goals (SDGs). to democratize problem-solving But, despite their promise, these The Initiative is open to all inter‑ with AI. projects face considerable chal‑ ested parties. lenges in gaining the scale neces‑ sary to achieve global impact.
Complete this form to express your interest. AI for Good Global Summit ITU News MAGAZINE 02/2020 19
The Initiative will offer assemblies A Workstream on “Repositories” 3 workstreams of resources to launch new AI will curate resources spanning projects aligned with the SDGs, from knowledge and AI capa‑ and scale them up fast. bilities to directories of experts, ongoing AI for Good projects, It will connect AI specialists, AI and initiatives developing rele‑ users, data owners and experts in vant standards and frameworks. domains to benefit from AI appli‑ It will support the connection of cations, helping them to align problem statements with projects incentives for innovation and pursuing solutions and identify Repositories develop AI solutions to precisely factors influencing project scala‑ 1 defined problems. AI develop‑ bility and sustainability. 2 Marketplaces ment and application will build on state-of-the-art infrastructure, A Workstream on “Marketplaces” 3 Project scaling enabling AI solutions to scale will enact a connection to mar‑ with the help of shared datasets, ketplaces of datasets and AI testing and simulation environ‑ models, capturing knowledge ments, AI models and associated supporting problem solving with software, and storage and com‑ AI. It will study incentives and puting resources. other considerations for contribu‑ tions of datasets and AI models to the Commons. “Project One” will support Three workstreams the growth of Technovation, A third Workstream on “Projects an initiative helping young The Initiative adopts a structure scaling” aims to scale up existing women and families to explore elaborated by a January 2020 projects and initiatives, support‑ core concepts of AI and apply roundtable welcoming some ing inclusive access to AI and AI tools to solve problems in 100 representatives of industry, data as tools to advance sustaina‑ their communities as part of academia, government, United ble development. global competitions. Nations agencies and technical standards bodies. “Project Zero” will work “Project Two” will support towards the establishment of a the new XPRIZE Pandemic It will build on standards and “global service platform” built Alliance in collecting the frameworks under develop‑ on the automated analysis data necessary to support ment globally, to the mutual of satellite imagery, opening globally coordinated benefit of their development access to the infrastructure responses to pandemics and implementation. required for associated AI such as COVID‑19. services to achieve immediate global reach. AI for Good goes digital ITU News MAGAZINE 02/2020 20 SeongJoon Cho/Bloomberg via Getty Images via Getty SeongJoon Cho/Bloomberg
COVID‑19: How Korea used innovative technology and AI to flatten the curve
By ITU News
J The Republic of Korea But there’s a lot more to managed to contain COVID‑19 the success. Information and without shutting down its communication economy, even though schools, ITU listened to experts, live on technologies are museums and gyms — places 27 March from the Republic of required to enhance where large numbers of people Korea, during Episode 1 of the traditional control might gather in an enclosed ITU AI for Good webinar series, measures and response space — were closed. who shared their experience and measures, as well as response to the pandemic, and for the development of How did they do that? how innovative technologies were innovative solutions. used to help flatten the curve in The country’s experience from the country. the previous SARS outbreak and Seon Kui Lee the more recent MERS (in 2015) “We still need to learn more Director of the Division of Risk enabled it to prepare in advance about [COVID‑19’s] characteris‑ Assessment and International to combat COVID‑19. tics — although we do know now Cooperation at KCDC AI for Good goes digital ITU News MAGAZINE 02/2020 21
that it is very contagious,” said “To develop a test in such a Seon Kui Lee, Director of the short time would not have been To develop a test in Division of Risk Assessment and possible without AI,” said Tai- such a short time International Cooperation at the Myoung Chung, Professor in would not have been Republic of Korea’s Center for the Department of Interaction possible without AI. Disease Control and Prevention Science at Sungkyunkwan (KCDC). “Information and com‑ University (SKKU).” munication technologies are Tai‑Myoung Chung required to enhance traditional The testing kit, which would Professor in the Department control measures and response normally have taken two to three of Interaction Science at SKKU measures, as well as for the devel‑ months to develop, was approved opment of innovative solutions.” by the authorities within less than a week of its application, and also Here are some of the top reasons certified in the European Union. for success: Just a few weeks later, when an Widespread testing in the enormous cluster of COVID‑19 Republic of Korea is targeted Fast-developed coronavirus cases emerged in mainly at the high-risk groups, i.e. testing kit the city of Daegu, it was ready those with underlying diseases, for testing. the elderly, people who share An important part homes, or live in crowded areas, of the Republic of The number of cases in the and passengers at arrival points Korea’s strategy since Republic of Korea peaked on emanating from countries with the outbreak of the COVID‑19 29 February, recording a total cases of the coronavirus or other coronavirus has been wide‑ of 909 infections. infectious diseases in the past spread testing. such as SARS or MERS. “Currently we have a total of Thanks to artificial intelligence 118 testing stations available (AI), the development of a nationwide, and have a testing Smart coronavirus testing kit happened capacity of 15 000 per day on quarantine fast in the country. Life sciences average — with a maximum of information company Seegene came up with 20 000 per day,” said Lee. system a coronavirus testing kit in under just three weeks. “It began with only lab testing, Dr Lee told the which later on expanded to local webinar audience that a quaran‑ governments and also to the tine information system was put primary medical laboratories and in place after the MERS outbreak hospitals,” said Dr Lee. in 2015. AI for Good goes digital ITU News MAGAZINE 02/2020 22
Mobile phone technology data for contact tracing
As well as inter‑ viewing, officials use location data from mobile phones, credit-card transaction records and CCTV 5 reasons why the Republic of Korea is footage to trace and test peo‑ a global leader in ICT ple who might have recently come into contact with an infected person.
In many places, detailed maps are published showing precise “Even before this COVID‑19 out‑ This enables frontline health movements of infected people, break, inbound travellers entering workers to have a full record of encouraging others who thought the Republic of Korea have been the patient’s history of move‑ they might have been in contact required to be checked for fever ments to help them quickly with an infected person to seek and to also to fill out a health identify and isolate or treat the out testing. questionnaire,” explains Dr Lee. suspected coronavirus patient in a timely manner. Inbound passengers with symp‑ AI for toms or having travelled to or In-bound travellers are required improving from a risk country are placed to download a self-health check diagnosis in quarantine. mobile app to their smartphone efficiency and and submit their health conditions patient “Information about the inbound on this app during their 14-day classification traveller from the Ministry of incubation period, explains Lee. Justice, the Ministry of Foreign Lee explained that as part of the Affairs, airline companies and With the collaboration of telecom risk-mitigation strategy, KCDC major telephone telecommunica‑ companies, they also receive texts established a system which tion companies are collected by messages and receive guidance categorized confirmed cases into KCDC’s quarantine information on how to report any COVID‑19 four categories: mild, moderate, system,” says Dr Lee. coronavirus symptoms they severe and very severe. might have developed while in quarantine. AI for Good goes digital ITU News MAGAZINE 02/2020 23
“Each category receives a differ‑ Mobile apps For example, one mobile app has ent treatment and is admitted to a for information been developed to direct those different facility according to the sharing who develop symptoms to the severity of the case,” said Dr Lee. nearest available testing station. Many mobile apps Another can indicate the nearest Professor Chung gave examples have been very point of purchase for availa‑ of AI‑based tools that are being quickly developed in the country ble masks. used in the country to enable the since the outbreak of the virus. quick diagnosis and classification A public chat robot using AI tech‑ of patients. They have proved useful for niques is being used to inform on information sharing to advise and ways of responding to coronavi‑ For example, VUNO’s Chest inform the public. “This would be rus, and another AI‑based voice X-Ray AI Image Support Decision a huge task for already stretched robot automatically calls people Tool — an algorithm for identify‑ health-care staff and volunteers,” who need attention, explained ing abnormal findings on chest claimed Professor Chung, saying Professor Chung. X‑rays — classifies intensive care that they can be developed patients by using X‑ray images quickly within a few days to and can examine the lung within a week. Daegu: just three seconds. Making use of a smart JLK inspection, though numer‑ city hub ous studies have developed an all-in-one medical platform called Daegu City, AiHub for disease diagnosis, where the majority of the which, it says, uses world class Republic of Korea’s cases of AI and big data technology from COVID‑19 have occurred so far, is various imaging devices. It can currently in the process of being examine lung disease within sec‑ transformed into a smart city (set onds using an AI technique that is Find out more about how the to be completed in 2021). being used in hospitals. Republic of Korea used ICTs and AI to flatten the curve on The epidemiological investigation the coronavirus by listening The company has also produced to the recording of Episode during the outbreak was able to an AI‑based, hand-held chest No. 1 of the AI for Good use the data hub of the smart city, X‑ray camera which can scan the Webinar Series. particularly for tracing patient chest in just three seconds and routes, which Chung says is “criti‑ The AI for Good Webinar give a heatmap visualization of cal to developing a new medicine Series is a free, live series of abnormal lesion. talks, interviews and panels, as quickly as possible.” featuring inter-disciplinary experts whose ideas, insights and solutions can help humanity leverage AI for good. AI for Good goes digital ITU News MAGAZINE 02/2020 24 Shutterstock
COVID‑19: Where are the self-driving cars and trucks?
By ITU News
J At a time when people around due to health concerns amid the of the Autonomous Drivers the world are observing social COVID‑19 pandemic, it appears Alliance (ADA), asked as he distancing measures to halt the we are still a long way from a fully moderated the fifth episode (held spread of the global COVID‑19 autonomous future. on 24 April) of the AI for Good pandemic, driverless mobility webinar series on COVID‑19: — which, by definition, offers Where are the self-driving cars mobility with minimal human Barriers to deployment and trucks? interaction — seems more relevant than ever. “2020 was billed as ’Prime Time’ The discussion looked at what for the self-driving car industry, barriers remain for autonomous Yet, as developers and research‑ so where are the self-driving cars vehicle deployment and how ers around the world have and trucks we were promised?” the future of self-driving vehicles suspended real-world testing of Bryn Balcombe, Chief Strategy may be reshaped following the their autonomous vehicle fleets Officer at Roborace and Founder COVID‑19 pandemic. AI for Good goes digital ITU News MAGAZINE 02/2020 25
One of the panellists, Michelle But in the wake of the current Avary, Head of Automotive pandemic, we need to expand Those companies and Autonomous Mobility at the definition of “safety” in the that survive will be the World Economic Forum, automotive sphere to include the ones that figure pointed to three core stumbling different types of safety such as out that slow-speed blocks that autonomous vehicles biosafety, said Avary. delivery operations face today. is the right way to go The panellists agreed that the in the near-term. “I think we need clarity on regu‑ COVID‑19 pandemic has high‑ lation globally — without a doubt, lighted a core business case for that’s needed. But we also need autonomous vehicle deployment Missy Cummings technological validity, as well as within the logistics and delivery Duke Pratt School of business model validity,” she said. space — certainly in the near-term. Engineering, US
“What we saw coming out of Slow moving deliveries — a China in using non-contact, low- potential business model? speed, highly-automated vehicles for delivery in areas that were A driverless future is often locked down, we really should And as the global pandemic mentioned in the same breath as explore more of these and think continues, the companies that can increased road safety, hailed as a about where we can remove pivot to this slow-speed delivery means to reduce the 1.35 million some of the biosafety risks to our of goods, such as groceries or global road fatalities every year. essential workers,” said Avary. medicine “have the best chance of surviving,” said panellist Missy Cummings, Professor in the Department of Electrical and Computer Engineering, Duke The COVID‑19 crisis has reignited debates on the Pratt School of Engineering, rapid deployment of autonomous vehicles United States.
“I think that the COVID‑19 sce‑ nario is going to really decimate the driverless car community,” she said. “Those companies that survive will be the ones that figure out that slow-speed delivery operations is the right way to go in the near-term.” AI for Good goes digital ITU News MAGAZINE 02/2020 26
There’s still a problem with prediction in Despite current trends away terms of predicting A ’vision test for cars’? from public transport systems — the behaviour of all notably in China, which has seen the other road users… Additionally, Cummings called for a dramatic drop in readership the global autonomous vehicle since the easing of the national community to establish “a vision COVID‑19 lockdown — panellists Bryn Balcombe test for cars” to continuously test saw this as the first step towards Chief Strategy Officer at and validate these technologies. mass autonomous transit systems Roborace and Founder of ADA in the longer-term, as a way to But any meaningful progress on meet current biosafety standards road safety will rely on global (i.e. social distancing measures) cooperation among the private and address ongoing trends in sector and governments, the on-demand transport. the world around them and act participants agreed. accordingly — are having to make the most reasonable guess about “We really need to make sure While people can which course of action to take that we’re sharing the data in the understand and predict under the circumstances. learning more widely, because we don’t think that every single com‑ But before this can become a “When the system doesn’t per‑ pany and every single operator reality, both for the COVID‑19 form the same way every time, needs to learn safety first-hand,” pandemic and beyond, the driv‑ even under the most benign said Avary. “There’s plenty of erless future faces safety issues of conditions, then how can we places to compete, but we don’t another kind: “perception”. ever put guarantees on its ability believe that core safety is one to navigate the world safely?” of them.” “There’s still a problem with pre‑ said Cummings. diction in terms of predicting the So in much the same way that behaviour of all the other road But it is not just a simple hardware the global community is working users and I think that’s actually fix, she added. “You are not going together to contain the COVID‑19 something that will be coming to solve this problem by putting pandemic through the sharing of to the fore as we progress in more sensors on a car, because best practices in network main‑ the perception quality system,” until we figure out how to at least tenance and collaborating to said Balcombe. replicate in part judgment under ensure #LearningNeverStops, the uncertainty, all the sensors in automotive industry must come While people can understand the world are not going to solve together to ensure the safe and and predict pedestrian intent, that problem.” steady roll out of autonomous autonomous perception systems vehicles — and increase road — image processing and machine One way to overcome this issue, safety in all its forms, agreed par‑ learning probabilistic reasoning suggested Avary, is to establish ticipants. algorithms that help autonomous global minimum safety guidelines driving systems understand for perception systems. AI for Good goes digital ITU News MAGAZINE 02/2020 27 Shutterstock
Tech startups join AI for Good Innovation Factory with live pitching
By ITU News
J At the firstAI for Good any one of the United Nations You’ve got to have Innovation Factory Live Pitching Sustainable Development problem and solution Session held on 15 May, tech Goals (SDGs). validation, and then startup entrepreneurs had product and market 4 minutes each to pitch their fit, so that customers artificial intelligence (AI) project 5 startup projects agree that there is remotely to a team of mentors. something they are Founder and CEO Mikela willing to pay for. During the webinar, 5 entre‑ Druckman, presented UK tech preneurs had the opportunity startup Grey Parrot, which offers to showcase their tech startup a computer vision for waste man‑ Stephen Ibaraki projects, whose solutions aim agement — targeting customers in Co-Chairman of ACM to accelerate progress to meet the waste management industry. Practitioner Board
The 2020 AI for Good Global Summit is being held online throughout the year. AI for Good goes digital ITU News MAGAZINE 02/2020 28
Iran Davar Ardalan, Founder and Tech for the deaf “You’ve got to have problem Chief Storytelling Officer pre‑ Startup EQ4ALL helps and solution validation, and deaf people have sented IVOW AI and its product then product and market fit, so equal accessibility to CultureGraph, an Enterprise API information, education and that customers agree that there Platform that enhances consumer communication. is something they are willing to engagement through the lens pay for,” said Stephen Ibaraki, of culture. Co-Chairman of the Association for Computing Machinery (ACM) Founder Javier Cardona, pre‑ Practitioner Board. sented the Latin-American tech startup 1DOC3 which provides Equally important is how good easy access to doctors in the and solid the team is, says Ibaraki, Spanish-speaking world. citing grit, talent, hard work, per‑ severance, optimism and continu‑ Co-founder and CSO Kevin Lee, ing drive as key elements for tech presented EQ4ALL. Supported by startup success. the Republic of Korea’s Ministry of Information and Communication Vera Futorjanski, Speaker, Advisor, Technologies, the tech startup Digital Platforms and Ecosystems has a mission to bring the Expert, believes that tech startups deaf to have equal accessibil‑ are the creative force that will be ity to information, education much needed to build whatever and communication. this new post-pandemic normality will look like. Founder Markus Pohl, pre‑ sented Child Growth Monitor, What it takes for startups Startups, says Futorjanski, “are a mobile app, released early to succeed agile and fast. We need this because of COVID‑19, that innovation going forward … in all measures and diagnoses children Before the entrepreneurs got sectors,” she says. for malnutrition. going, however, the session began with mentors sharing Futorjanski is sure that working Each project presentation was fol‑ insightful tips on what it takes from home is here to stay, and lowed by a Q&A, which gave the to be a tech entrepreneur — says that “it will be a tough time opportunity for the tech startup and opportunity-rich areas for startups, as there might not be entrepreneurs to receive advice for innovation, such as AI and as much cash available.” and feedback from the mentors. digital health. AI for Good goes digital ITU News MAGAZINE 02/2020 29
The silver lining is The real value Big data has been that you can find is actually in a tremendous help talent easier — and innovation. to public health. talent is an important core part of building an ecosystem. Neil Sahota Zhi Zhen Qin Worldwide Business Senior Technical Officer at Development Leader the United Nations Office Vera Futorjanski at IBM Watson for Project Services Speaker, Advisor, Digital Platforms and Ecosystems Expert
On the other hand, she is con‑ This is the core trait of a startup in “Think about how much is going fident that it is a great time this time of uncertainty.” to change in the next 10 years — to build businesses, and that in our personal and professional downturns are a good time to lives,” says Sahota. “We have start companies. AI and innovation already seen amazing changes. If we were to jump forward to “The silver lining is that you can Neil Sahota, Worldwide Business 2030 and look back, we would be find talent easier — and talent Development Leader at IBM amazed to see how far we have is an important core part of Watson and author of the book actually come in a short time.” building an ecosystem,” she told “AI Revolution,” shared his the audience. thoughts on AI and innovation. AI, Big Data and digital Futorjanski believes that govern‑ “AI is going to be a revolutionary health solutions ments will realize how important force, because it is going to feed those ecosystems and tech start‑ the Fourth Industrial Revolution,” Zhi Zhen Qin, Senior Technical ups are. he says. Officer at the United Nations Office for Project Services, spoke “I’m optimistic: I think we will see It is not about automation, says about how traditional public more support [for startups, from Sahota, explaining that auto‑ health control methods have lim‑ governments] going forward,” mation is all about improving itations when it comes to tackling she says. “As Charles Darwin something that you already have. COVID‑19. The potential of digital said: ’Those who survive are the “The real value is actually in inno‑ technology, says Qin, “can have a ones most adaptable to change.’ vation,” he says. huge benefit in augmenting tradi‑ tional public health strategies.” AI for Good goes digital ITU News MAGAZINE 02/2020 30
Quin highlights the usefulness of Key challenges to “Patients are concerned,” she says, real-time data tracking. “We see scaling up “and they will likely want care in a countries adapting their national place where they are most com‑ strategy with data that they are Sasha Cahill, Advisor at IBM fortable — and that is going to be updating every day,” she says. Cloud Hyper Protect Accelerator their home setting.” highlights the most important The power of AI has a role to play challenges that tech startups “I want to challenge entre‑ in predicting the prognosis of need to overcome to actually preneurs to think about those COVID‑19 and to detect other scale up. impacts, and how we can think infectious diseases — not to men‑ about models that are scalable,” tion, chest X-rays and CT scans Cahill agrees with other mentors said Cahill. from hospitals around the world, that the technology has to be said Qin. needed — and the team has to be the right one — for a tech startup Pitch your startup! “Big data has been a tremendous to succeed. help to public health — there is a The Innovation Factory Live lot of big data analysis,” says Qin, Understandably, during this crisis, Pitching Sessions will continue to adding that without big data it many new tech startups are in the showcase a diverse set of emerg‑ is not possible to model disease healthcare sector. ing entrepreneurs with promising activity, predict potential growth, AI ventures to tackle our world’s perform strategic analyses, or Cahill also shared insights specific greatest challenges, and will measure population impact. to technology for healthcare. include discussions with featured mentors on topics related to Qin is sure that the healthcare “In the United States, physician social impact. industry will see big changes in time is limited to 10–15 mins the future, not just for COVID‑19, per patient. If the technology is Do you have an innovative idea but also for other diseases and too complex to use, then both or a tech startup using AI to help health care in general. The physician and patient will be achieve the SDGs? healthcare industry, she says, “has frustrated,” says Cahill. “How been compelled to adopt digi‑ can we as entrepreneurs create If so, then why not sign up for tal technology.” new delivery models which can a chance to pitch in the ITU AI be monetized?” for Good Innovation Factory? Applications are open until Cahill advises budding health 30 June. Apply here. care startups to think carefully about the patient experience. AI for sustainable development ITU News MAGAZINE 02/2020 31 unsplash
Africa’s AI community set to light up the AI for Good Global Summit
By Alexander Tsado and Nick Bradshaw
Alexander Tsado, Advisor for Alliance4AI and Nick Bradshaw, Co-Founder of AI Expo Africa
J It is becoming clearer to the Africa and AI world that we can only achieve For Africa, AI represents the United Nations Sustainable access, efficiency and Africa needs to adopt artificial Development Goals (SDGs) by a potential antidote to intelligence (AI) now for two 2030 if the innovative voices of unemployment and key reasons. those in Africa, who are most underdevelopment. affected by the challenges, Firstly, by 2030, it’ll be home to are involved. 46 per cent of the world’s youth. Alexander Tsado/ The continent’s rapid population This is a clarion call for the rise of Nick Bradshaw growth means we have to think Africa’s innovative voice for the Advisor for Alliance4AI/ critically about the future of work. fourth industrial revolution. Co‑Founder of AI Expo Africa AI for sustainable development ITU News MAGAZINE 02/2020 32
AI for claims and fraud detection in Africa Secondly, its annual infrastructure spend gap is USD 100 billion. Curacel’s AI platform enables insurers track fraud and automate It can benefit immensely from claims seamlessly. lowering these costs with transfor‑ mational technologies like AI.
For Africa, AI represents access, efficiency and a potential antidote to unemployment and underdevelopment.
To maximize the effectiveness of limited resources, every organi‑ zation and government seeking to meet its population’s needs should consider the power of AI to render their efforts more likely to succeed.
For example, governments can distribute the limited fertilizers Alliance4ai, AI Media group, AI for Good Global they purchase to the exact farm‑ AI for Development and Data Summit 2020 ers that need them or distribute Science Nigeria are some of the medical drugs to the right patient key private organizations pooling Alliance4ai and the AI Media populations. Companies with together resources to galvanize Group are partnering to show‑ small budgets can adequately interest and activate African case at the summit globally market to the right customers, luminaries solving problems with relevant solutions built in Africa sell more, grow their business the technology. by Africa’s top innovators. The top to finally be able to create jobs highlights include: for Africans. So far, these organizations reached over 25 000 Africans In addition, Africa is perhaps in 2019. the continent with the greatest diversity of data and use cases to feed the design of AI solutions that can be applied in most parts of the world. AI for sustainable development ITU News MAGAZINE 02/2020 33
WeFly Agri for monitoring plantations Keynote addresses speaking to the importance of Africa’s WeFly Agri is an Ivorian Start Up that enables farmers to view and Innovative voice that seek to remotely monitor their plantation activities on site. drive inclusion in AI.
Talks on data center and supercomputer projects being developed in Africa to power world-class innovation.
Learnings and insights from top AI startups and AI training communities in Africa.
Demonstrations from top researchers addressing critical issues such as fight against Malaria, preserving African languages and culture and creating a template of innovation and collaboration For us, the future couldn’t A successful and empowered for other low-resource be brighter. Africa is one that is an exporter of languages to be represented. knowledge and solutions to the At the AI for Good Global Summit world, and we will all be better we will together share our solu‑ for it. tions, form partnerships and col‑ laborate as we collectively strive to achieve the 2030 sustainable development goals. AI for sustainable development ITU News MAGAZINE 02/2020 34 Shutterstock
How can we solve the problems of gender bias in AI? Experts weigh in.
By ITU News
J Artificial intelligence (AI) has a If AI is to help accelerate pro‑ bias problem. gress on humanity’s greatest challenges, we will need to work The problem with Actually, AI has many well-doc‑ together to solve AI’s many gen‑ lack of diversity umented bias problems. Chief der bias problems. is that you have among these, arguably, is gen‑ blind sides. der bias. But how?
From the creation of data sets, to What equitable policies need Ida Tin the way the data is collected and to be made around AI? Why is CEO and Co-Founder of Clue used, to the building of AI solu‑ diversity so important for data tions — women are underrepre‑ collection? What does algorithmic sented at every stage. This means gender bias look like? How can that AI solutions will not serve the AI show us COVID‑19’s impact on needs of half the world. roles in the workplace?
Note: Due to the global COVID‑19 pandemic, this year’s AI for Good Global Summit is all online. AI for sustainable development ITU News MAGAZINE 02/2020 35
A group of experts gathered “We are well beyond the time on 5 May 2020 at the open‑ Why would the people where we have to include the ing “Breakthrough” session of you’re building solutions people that we are building the the annual AI for Good Global for not be involved solutions for,” she said. Summit to tackle these questions in the process? and more. How does gender Kishau Rogers inequality in AI begin? Including women at CEO of Time Study every stage “Everything begins with data sets,” said Kaitlin Kraft-Buchman, A large part of the problem is that Founder and CEO of Women@ AI solutions are not being made TheTable. “Data sets have left with women in mind, agreed the out women.” panel during the webinar. “A huge part of our need space Then, algorithms get skewed and “When you think about making was missed, because those who machine learning exacerbates something that is truly valuable built products weren’t women” these problems, she added, for people, you have to think said Ms Tin. “We desperately providing many examples of bias about how it’s helpful for every need diverse teams asking these against women in data, algo‑ individual,” said Ida Tin, CEO and questions. It’s fundamental. If not, rithms and machine learning — Co-Founder of Clue, a custom‑ you end up building a world that’s from banking to the job market, ized female health period-track‑ not for everybody.” and beyond. ing app with 12 million active users worldwide. The panelists agreed that the “Data tells a story. It’s more stakes are high with AI and that than just fields and values,” said “The problem with lack of diver‑ the mistakes of the past must not Ms Rogers. “My first step is to sity is that you have blind sides,” be repeated. understand the data — how did it said Ms Tin. “There is a profound get to this place?” blindness in the world that … is “Why would the people you’re embedded in culture.” building solutions for not be We need to look at open, collab‑ involved in the process?” asked orative data sets and we need to Men often want to understand the Kishau Rogers, CEO of Time reevaluate what’s in the original issues women face, she said, but Study, a startup offering solu‑ data, said Ms Kraft-Buchman. how would they know what those tions for using machine learning, “That’s important for policy mak‑ are? How would they even think advanced natural language ers,” she added. to ask? processing, and data science to automatically tell a story of how enterprise employees spend their time. AI for sustainable development ITU News MAGAZINE 02/2020 36
Bringing ethics to What can be done on the The Future of Us, the discussion policy side? AI + Gender
Panelists agreed on the impor‑ The public sector should consider tance of ethics in this discussion. setting aside funding for wom‑ en-owned tech businesses with “If you’re going to release a women-run teams of developers, new system into the world, you said Kraft-Buchman. need to think very hard about your social responsibility,” said “It’s not about creating a bunch Andy Coravos, CEO/Founder of of female Zuckerbergs,” she said, This breakthrough session addresses current gender Elektra Labs. adding that women need to be issues in AI, including working throughout the ranks but not limited to Kraft-Buchman mentioned that of tech companies building the algorithmic bias and data there should be a Hippocratic latest AI solutions. “It’s not only inclusivity. The goal of this session is to identify AI Oath for AI, similar to the one about making more unicorns or solutions that empower used in the medical profession to owning the company. We need to underrepresented promise to work for the benefit go broad and wide.” communities and enable of people. an equitable future for humanity. “Policy makers need to have “The challenge with Hypcratic courage — and be encouraged — Oaths is that people don’t to ask questions about how the remember them,” responded technology is made,” she said. Ms. Coravos. “What’s better are checklists,” she said, giving She also called for “a few less examples of professions, such as pizza-delivery apps” — and a few “What I’d like to see is a broader pilots, where checklists must be more solutions that are using AI vision for who makes AI,” said followed in order to operate. for social good. Rogers. “You have to include people every step of the way. Ethics should be core and fun‑ There are so many big problems, We have to put the human in the damental to technical classes but we are “somehow lacking the centre of this entire process. We for those learning how to cre‑ imagination to use tech to solve don’t have an algorithm crisis we ate tomorrow’s AI solutions, these problems,” she said. “Let’s have a crisis of caring. We have to offered Rogers. use the tech to correct these.” care more about the people we’re creating solutions for.” “We need a really deep conversa‑ In the end, it’s not a technical tion about ethics and about what problem, it’s a human problem, world we actually want to create,” agreed panelists. said Tin. AI for sustainable development ITU News MAGAZINE 02/2020 37 Shutterstock
COVID‑19: How can AI help fight hunger and food insecurity?
By ITU News
J For decades, the number of The World Food Programme’s hungry people was declining — 2020 Global Report on Food This will interact this isn’t true anymore. Crises describes the scale of with a reverse shift acute hunger in the world. It pro‑ in globalization to The coronavirus pandemic is vides an analysis of the drivers more decentralized having an unprecedented impact that are contributing to food cri‑ food systems. on food systems. But even before ses across the globe, and exam‑ the COVID‑19 pandemic began ines how the COVID‑19 pandemic to affect food systems and agri‑ might contribute to their perpetu‑ Lorin Fries cultural livelihoods, 135 million ation or deterioration. FutureTable people were already facing acute food insecurity.
Note: This year’s AI for Good Global Summit is being held online throughout the year. Don’t miss the next online discussion. AI for sustainable development ITU News MAGAZINE 02/2020 38
Experts came together in a The Future of robotics and automation,” said webinar “Breakthrough” session Food, AI + Food Fries, also indicating that “the of the annual AI for Good Global question of the future of work Summit, on 7 May, to discuss food for workers in this turbo-charged insecurity and possible solutions. digital age is an open one.” Moderated by Caroline Kolta, Senior Associate of XPrize, the Entrepreneurs and small and panel explored how AI and inno‑ medium-size enterprises (SMEs) vative technology can help shape are experiencing high impact the food landscape of tomorrow. from the current food crisis. Here’s what they had to say. This breakthrough session SMEs in Africa, are considered to tackled issues around food be the backbone of the economy, insecurity, including but not limited to the impact providing food services to their Systemic change — a of COVID‑19 on food communities. “How we can sup‑ window of opportunity systems. The goal of this port the SMEs is then an imper‑ session was to identify how ative question at this moment?” AI can help us reach zero Lorin Fries, Founder and Lead hunger and shape the food said Fries. Partner, FutureTable, highlighted landscape of tomorrow. some of the hardest hit actors by COVID‑19 and suggested ways AI innovative solutions that the food crisis might bring not only change, but a window AI is already helping in agricul‑ of opportunity. ture; for example, picking vegeta‑ Organization. It shows what might bles, controlling pest infestations, Those from import-dependent be required for the successful soil and crop health monitoring nations and regions, explained implementation of decentralized and predictive analysis. Fries, would be some of the hard‑ food systems. Information and est hit — Africa for example — and communication technologies Bernard Kowatsch, Head of small-island developing states. (ICTs) could be part of that. Innovation Accelerator, United Africa is already set to import Nations World Food Programme 110 billion USD of food by 2025. Fries points out that while we (WFP), the world’s leading agency “This will interact with a reverse are recognizing the essential fighting hunger, outlined some shift in globalization to more role of farm, factory and food of WFP’s innovative solutions to decentralized food systems,” workers, at the same time we food insecurity. said Fries. are realizing that any human interaction with the food value Within days of the pandemic A case study on a decentralized chain, at least in the context of a being declared, HungerMap food system in Mozambique virus, presents some risk. “So you LIVE was tracking coronavi‑ was published by the United see a lot of activity in the invest‑ rus outbreaks in most low and Nations Food and Agriculture ment space for instance around lower-middle income countries. AI for sustainable development ITU News MAGAZINE 02/2020 39
WFP’s Hunger Monitoring unit is Reducing food waste An app in Nigeria is helping to reduce waste in the food system. now collecting, analysing and vis‑ ualizing real-time data to scale up remote food security monitoring to 40 countries. The map enables a quicker response to emergen‑ cies. “You can be faster in getting the right food — at the right time. And that can have a huge impact”, says Kowatsch.
By making it easy to analyse connected data streams, WFP is helping the humanitarian community make fast, evi‑ dence-based decisions.
Another WPF innovation is the PLUS School Menu Optimizer Fries sees producing food else‑ An app in Nigeria is helping to tool. The tool is an online software where such as indoor farming reduce waste in the food system that, through an advanced math‑ accelerating forward in future. (see video). ematical algorithm, calculates the most nutritious and cost- effective “Supply chain logistics and the school feeding menus to ensure efficiencies that can be gained From ’linear’ to ’circular’ dietary diversity based on local through digitization and some food system food where possible. of the AI apps will be absolutely key,” says Fries. Improving trace‑ Merijn Dols, Global Director of The WFP e‑shop Somalia is ena‑ ability in food value chains is one Open Innovation and Circular bling citizens to shop comfortably key requirement. Economy of Food, Danone antic‑ from any authorized World Food ipates some opportunity during Programme retailer in the country. E‑commerce is already shap‑ this food crisis, to re-design ing how people buy food, from and innovate. Kowatsch also cited Hello Tractor, Alibaba and Amazon down a startup which claims to offer to marketplaces, but Fries is “This unprecedented shock to the technology for smarter, better convinced that there is a set of system is making the downsides maintained, and more profitable innovative market spaces such of our linear food system that we tractors now powered by artifi‑ as Nigeria, that are doing a have created very visible, and cial intelligence for better and better job of linking supply and even undeniable,” said Dols. “And faster decision-making. demand, and balancing demand this might be the start of an accel‑ and supply. eration in the system shift.” AI for sustainable development ITU News MAGAZINE 02/2020 40
Dols explained that our current “This is where particularly local linear food system is built on AI can have a positive talent could be supportive — in a view of the world of a giant impact on a different, some of the hot spots in Africa machine driving growth, where regenerative future. and the Middle East, and some one of the key levers for this other areas, across Asia and Latin growth, and the driver of the America,” said Kowatsch. “There system, is efficiency. Merijn Dols is talent, and interesting start-ups, Danone and definitely drive.” “Those start- Hence, Dols strongly advocates ups have better knowledge of for a circular food system — one the local innovation ecosystem,” that has positive economic and he says. social impact by design. A sys‑ tem which, he explains, aims at Referring to the AI hotspots; dynamic balance rather than Silicon Valley, some parts of infinite growth. “A system that is “AI can have a positive impact on Europe, China and other places not driven by efficiency, but it is a different, regenerative future,” with lots of AI researchers, driven by effectiveness. Which enthuses Dols. Kowatsch questioned how that is about doing the right thing,” technology can be provided and says Dols. made more accessible to other AI in developing countries parts of the world. Could compa‑ According to Dols, building a cir‑ — benefiting those who nies provide pro bono resources, cular system of food would need need it most or focus on building up local AI. With highly local farming, for innovation hubs in Africa? “Some example, farmers will need data How does food investment in AI of the tech companies are doing that is relevant to their crops, and in developing countries look like? that.” he says. to identify the crops that work best in a specific environment. Bernard Kowatsch is of the opin‑ Kowatsch suggested that there is AI can be harnessed for connect‑ ion that venture capital money is a need to inspire people — serial ing producers and consumers. generally underrepresented in entrepreneurs who can actu‑ food compared to other sectors. ally create a business that has a “I see AI helping producers and Even there, he says a lot of the positive impact on the planet — in consumers connect real-time.” investment goes to food delivery ending hunger. says Dols, adding that “real-time apps, while there is potential for access to holistic, understandable much to be done in develop‑ data will help everyone.” ing countries. AI for sustainable development ITU News MAGAZINE 02/2020 41
Hello Tractor Connecting you to your tractor and your tractor to the world She warned against the costs of using AI, such as for exam‑ You can create a Hello Tractor offers start-up and create ple, the potential perpetua‑ technology for smarter, jobs for yourself and tion of bias, and the need for better maintained, and others and still have clear re-designing. more profitable tractors now powered by artificial a positive impact. intelligence for better and “How do we actually put ethics faster decision making. and equity at the centre of our AI Bernard Kowatsch systems? asked Fries, “including How does it work? World Food Programme and especially as applied to our food systems?”
She also cautioned how “data is the new gold, and much of it is owned, and will be owned by large players — big businesses.” “This is a huge opportunity… You can create a start-up and She encouraged reflection on Farmer requests tractor create jobs for yourself and others how the data will be accessed by service via app or booking and still have a positive impact.” the people who use it; the farm‑ agents ers, for example, and consumers. “We need to broadcast this more and be open to it,” said Kowatsch. “We need more open source systems, more protections, more savvy citizens and more tech The costs of AI knowledgeable consumers,” said Fries. “It is complicated, and col‑ For Lorin Fries, collaboration laboration is very difficult. Actors Tractor owner pairs request is critical. “There is a world of speak different languages — and with available tractor opportunity at the intersection have different cultures.” of technology like AI and issues like hunger or social impact more The panel agreed, nevertheless, broadly,” she says, adding that that having people at the centre, “we need a multi-stakeholder to help transform AI for social approach. Business to come good will be key in helping to alongside governments — along‑ fight hunger and food insecurity, side civil society — alongside a and reshaping the food systems range of players involved.” of tomorrow. Farmer is served and tractor is monitored
Source: Hello Tractor Site AI for sustainable development ITU News MAGAZINE 02/2020 42 Shutterstock
How AI can help save the planet
By ITU News
J Fish have returned to the This is where artificial intelligence canals of Venice. Villagers (AI) can help. across northern India can see the Himalayas over 200 Km The AI for Good Global Summit We’re finally at the away. Cities around the world 2020 breakthrough session on point where we can are recording the cleanest air AI and the environment (held on use remote sensing on record. 8 May), looked at how AI solu‑ and AI to make tions can shed light on environ‑ climate change As the world recently celebrated mentally destructive practices more concrete. the 50th anniversary of Earth Day, as well as protect our planet. COVID‑19 lockdowns have shown Panellists presented compelling Sasha Luccioni us a glimpse of what a world with‑ and interesting applications of AI out pollution could look like. But and machine learning (ML) to not Quebec Artificial Intelligence Institute staying home is not a long-term just observe Earth systems, but solution to climate action — and as actually put insight into practi‑ we slowly return to normal, so too cal action. will pollution levels. AI for sustainable development ITU News MAGAZINE 02/2020 43
Fixing something invisible The Future Intelligent and of Earth, AI + actionable information It might not be difficult to imagine The Environment what London looks like under For Andrew Zolli, Vice President of 10 metres of water given the Global Impact and Planet Fellow proclivities of British weather, but at Planet Labs, visualization is just what about Montreal or Tokyo? one piece of the puzzle; making data and intelligence into actiona‑ “I find that the main obstacle ble insights for policymakers and for humanity to act on climate local leaders is largely a question change is the problem we of accessibility. have of actually imagining it This breakthrough session explored how AI solutions and visualizing climate change, “We spend a lot of time talking can shed light on and I think that AI can help environmentally about visualization because with that,” said Sasha Luccioni, destructive practices as we’re trying to prompt people Director of Scientific Projects well as protect our world. to urgency. But the reality is, for The goal of the session and Postdoctoral Researcher utility, what you really want to do was to take urgent action at Mila — Quebec Artificial in combatting climate is get the answer down to like Intelligence Institute. change, to identify a one, or a zero. Is there a thing solutions to sustainably there? Is that deforestation or use our planet’s resources, She is developing a visualiza‑ not?” he said. and to re-imagine the tion tool using existing open- future of the Earth with AI. source climate models to make Using a constellation of satellites, the potential consequences of Planet Labs takes images of the climate change visible, accessible entire surface of the Earth every and actionable, to help people to day at roughly three metres per understand both what is and what pixel. A second group of satel‑ might happen in the future. lites can then zoom in selectively anywhere on Earth, and image “Our idea is to help people really Drawing parallels with the impact multiple times a day at 70 to realize what this means with a of images of the hole in the 80 centimetres per pixel. “We use 1.5-degree climate increase,” Ozone layer which played a big the daily monitoring of the Earth she said. part of the signing of the Montreal to understand what gross level Protocol, “we’re finally at the point changes are occurring,” he said. where we can use remote sensing and AI to make climate change more concrete,” she said. AI for sustainable development ITU News MAGAZINE 02/2020 44
Planet data mapping “We can’t fix what we can’t see,” Zolli said. Planet provides custom monitoring, tasking, and basemaps for mapping and geographic information system (GIS) applications. It is also being used to monitor and predict agricultural output to Morning fog prevent potential famine. Parque Nacional Bernardo O’Higgins, “Right now, when, especially Chile in East Africa, we’re seeing the Fog flowed through the intersection of COVID‑19 plus steep valleys and fjords of the locust infestation creating real southern Chile on March 28, 2020. potential risk and vulnerability. PlanetScope These kinds of tools will be used Rock avalanche in a very actionable way — in fact, on Salkantay they already are — to help deter‑ Vilcabamba Range, Peru mine how to keep people safe On February 23, 2020, a and make sure that they get fed,” massive flood of mud and said Zolli. rocks flowed down the Salkantay River in Peru.
Skysat Direct impact on Icefields on the environment Kilimanjaro Mount Kilimanjaro, One key environmental ques‑ Tanzania tion that panellists addressed This SkySat image shows was about the technology
Planet Gallery the remaining ice fields on itself: Isn’t the immense com‑ the summit of Africa’s tallest peak, Mount Kilimanjaro. puting power needed for deep Skysat
Source: Source: learning counterproductive for the environment?
While there are bigger models, especially in natural language The data is being used to monitor protection, that for example, processing, that can produce a
urban growth and predict where show us the earliest signs of coral lot of CO2, “the day-to-day, more populations are exposed to risks, bleaching to help de-intensify small-scale AI efforts definitely such as flooding; build moni‑ human impacts, and track the car‑ don’t produce that much,” toring systems for biodiversity bon emissions of coal fired power said Luccioni. conservation and an ecological plants around the world. AI for sustainable development ITU News MAGAZINE 02/2020 45
“One of the things to see is at But it is not just a case of having One of the things an industrial scale — lowering the technology; implementing to see is at an the cost, improving the energy AI solutions relies on data — industrial scale — efficiency of AI — because both something that on-the-ground lowering the cost, renewable solar energy and the non-governmental organizations improving the energy computing architectures them‑ (NGOs) may not have ready efficiency of AI. selves are becoming broadly access to. speaking more energy efficient and more powerful over time,” After approaching NGOs in Andrew Zolli he said. Montreal to help them develop Global Impact and Planet AI-led solutions, she realized that Fellow at Planet Labs there was a fundamental flaw in An AI-enabled future for her plan: “Most of the answers the environment we got were: ’what data? Can you give us lessons on how to use The panellists were optimis‑ Excel, or how to set up a manage‑ tic about the future of an ment system for our inventory?’” AI-enabled environment. She is creating a downloada‑ As such, Luccioni advocated ble tool that tracks and calcu‑ “This is the foundational insight for close working partnerships lates machine learning energy that can be used for new forms between fieldworkers and consumption and translates of climate risk insurance or new AI experts in order to create this into CO equivalents to 2 realms of climate risk financing, or AI-led solutions. raise awareness. even for communities to advocate for their own climate rights and “Maybe five years down the road, Zolli pointed to larger efforts to for climate justice,” Zolli said. we’ll be ready for that, but now, improve the energy efficiency of we should start doing tutorials data centres and compute cen‑ Luccioni agreed: “I’ve always about what data is and how it can tres using AI — highlighting that seen AI and machine learning help,” she said. Google’s DeepMind successfully as another frontier… for actually reduced their energy consump‑ developing new solutions that tion for cooling their data centres can take us forward to that sus‑ by 40 per cent. tainable future.” AI for sustainable development ITU News MAGAZINE 02/2020 46 Unsplash
What the United Nations is doing to leverage AI for Good
By ITU Secretary-General Houlin Zhao
J Very few topics have captured responding to humanitarian crises the imagination of the public to tackling climate change. What is clear is that and promised to transform no nation, company humanity more than artificial Although it is still early days for AI, or organization intelligence (AI). the agencies highlight the need can meet these to improve data literacy skills challenges alone. In the publication “United Nations across their organizations and Activities on Artificial Intelligence”, warn against the capability gap more than 30 United Nations (UN) between developed and less-de‑ Houlin Zhao agencies and bodies describe veloped countries. They call for how they are increasingly using responsible innovation and a bet‑ ITU Secretary-General AI to meet many of the world’s ter understanding of the implica‑ most urgent challenges, from tions and potential benefits of AI.
This article is a Foreword to the compendium entitled, “United Nations Activities on Artificial Intelligence.“ AI for sustainable development ITU News MAGAZINE 02/2020 47
For AI to be a force for good, of AI to accelerate progress United Nations it must face complex questions towards meeting the United Activities about trust and address chal‑ Nations Sustainable Development on Artificial lenges ranging from job displace‑ Goals. This year’s event is being Intelligence (AI) ment to autonomous weapons held online, throughout the year. and potential bias in algorithms. What is clear is that no nation, The UN system has an important company or organization can role to play in balancing tech‑ meet these challenges alone. nological progress with social progress and ITU, as the UN spe‑ Each year, ITU’s AI for Good cialized agency for ICTs, is at the Global Summit brings thousands forefront of this critical endeavour. of AI innovators and public and private-sector decision makers I commend this publication to all from around the world, including those committed to ensuring that UN agencies, who come together artificial intelligence benefits all to identify practical applications of humanity.
A joint-effort between ITU and over 35 other UN agencies and bodies, and all partners of 2019’s AI for Good Global Summit, this report provides information about the diverse and innovative activities related to artificial intelligence (AI) across the UN system to meet many of our world’s increasingly urgent challenges, from responding to humanitarian crises to tackling climate change.
Read the report. Facing COVID-19 is one of humanity’s greatest challenges ever, bigger than any one government or organization can overcome without uncommon thinking and partnerships.
In response to this and future health crisis, XPRIZE, in collaboration with Anthem, launched the XPRIZE Pandemic Alliance, a global coalition of partners, including expert innovators, clinicians, researchers, data scientists, and health institutions. This alliance is moving toward predicting, diagnosing, and treating COVID-19, and preventing future pandemics by working together, sharing ideas, identifying areas of need, and offering resources. It will create a series of incentivized data challenges to help predict and prevent the next pandemic.
Alliance members may also participate in the XPRIZE Data Collaborative, a unique platform for researchers and innovators to exchange ideas and share and experiment with ethically sourced and validated data from a broad spectrum of fields in their search for solutions.
As people and businesses all over the world are doing what they can to stop the spread of COVID-19, fight the disease, heal the afflicted, safeguard those on the front lines, and mitigate JOIN A DATA- the economic fallout of this crisis, we can accelerate these critical response efforts and POWERED identify breakthrough solutions. Each day the rate of infection reaches new ALLIANCE heights, the economy contracts further, and the number of fatalities, people near and dear to us among them, continues to grow. Now is the time TO STOP to diminish the fear, to ignite unprecedented innovation, and stimulate exponential impact to COVID-19 bring humanity a healthy more resilient future. Here are several ways to help:
DATA EXPERTISE TECHNOLOGY
Provide access to critical Connect domain experts, Share code, algorithms, data to address the scientists, clinicians, data APIs, unique models, ongoing challenges of the scientists, and others cloud compute, and other pandemic. to data sets; inform the resources. challenge design.
AMPLIFICATION FUNDING
Promote the XPRIZE Sponsor a data challenge Pandemic Alliance to uncover insights and to communities of unlock solutions. innovators, medical researchers, and experts.
Join the XPRIZE Pandemic Alliance and help make a pandemic-free future.
covid19.xprize.org AI and health care ITU News MAGAZINE 02/2020 50 Luis Enrique Ascui/Getty Images Ascui/Getty Enrique Luis
Reducing the global healthcare shortfall using AI
By ITU News
J Dr Nick Sireau finally found — so he went to find one. After a The boom of digital a way to save his children from slow and expensive process, he health tech decades of disabling pain. found it. It is predicted that the healthcare Dr Sireau’s two sons were born But Dr Sireau’s story isn’t unique. landscape will completely shift with Alkaptonuria, also known as There are 7000 known rare dis‑ in the next 10 years, driven by AI Black Bone Disease, an ultra-rare eases which affect some 400 mil‑ and machine learning (AI/ML). By genetic mutation which affects lion people worldwide. However, aggregating and analysing data just 200 people worldwide. Left 95 per cent of these diseases do from connected-home devices untreated, they would have been not have an approved treatment, and medical records, healthcare expected to develop severe ear‑ and not everyone has the time systems will be able to deliver ly-onset osteoarthritis and have and resources available to find a proactive and predictive medi‑ a significantly heightened risk of treatment themselves. cal care. brittle bones and heart disease. This is where artificial intelligence But there was no medical treat‑ (AI) can help. ment available on the market AI and health care ITU News MAGAZINE 02/2020 51
“Health care has gotten too and make predictions about what complex for humans to do a good diseases people are likely to have We can find a disease job; there are way too many dis‑ and what treatments are likely to where there is no eases, way too many treatments — be most effective,” he said. treatment, crunch certainly if you factor in anything the data and make about genetics, or proteomics, In 2018, investments in health predictions… We’ve had or biomedics, it’s just way too tech surpassed USD 8.1 billion, a 100 per cent success much data for humans to be able with AI and machine learning rate so far, which to comprehend,” said Professor driving over a quarter of the we didn’t expect. Tom Davenport, President’s funds. Companies are testing Distinguished Professor of a wide range of AI healthcare Information Technology and solutions, from drug discovery Tim Guilliams Management at Babson College, to diagnostics — and the results Healx Wellesley, USA, told ITU in are promising. an interview.
“AI, particularly machine learning, Drug discovery for can deal with all of that data in rare diseases a relatively straightforward form 12 years vs. 24 months Dr Sireau’s story was the inspira‑ tion behind Healx, a healthcare Using this method, the company startup that uses AI to predict aims to discover new treatments and find treatments for rare dis‑ for rare diseases and move them eases at scale using existing and towards clinical trials within AI, particularly approved medical treatments. 24 months. This is a lot quicker machine learning, than the traditional model of dis‑ can deal with all of To analyse a disease, Healx runs covery and clinical development that data in a relatively algorithms that look through the of new medicines which takes straightforward form available data, such as clinical trial an average of 12 years and costs and make predictions information, the models available an estimated USD 2.6 billion. about what diseases to test the treatments, the number Moreover, only 12 per cent of people are likely and quality of patient groups and new medicines brought through to have and what how proactive they are. On the medical trial are approved. treatments are likely medication matching side, it’s the to be most effective. same thing; algorithms search “We can find a disease where and predict which existing med‑ there is no treatment, crunch ication is best suited to stop or the data and make predictions,” Professor Tom Davenport reverse the disease. Guilliams said. “We’ve had a Babson College, Wellesley, USA 100 per cent success rate so far, which we didn’t expect.” AI and health care ITU News MAGAZINE 02/2020 52
Working with medication that has Dan Vahdat, founder and CEO of Huma, speaks already been approved reduces to Proactive London’s Andrew Scott following its regulatory hurdles and speeds up rebranding from Medopad. the time to trial, making it possi‑ ble for a treatment to happen in a couple of years, as opposed to a couple of decades.
“For Fragile X Syndrome, it took us 18 months to make the predic‑ tions, test them, get them ready for phase 2a clinical trial at a frac‑ tion of the cost,” Guilliams said.
The start-up aims to help get 100 rare disease treatments towards clinical trial by 2025. They are currently focusing on two disease clusters, rare neuro‑ logical diseases and rare cancers, and plan to expand to a further For instance, Medopad, an Parkinson’s affects around 10 mil‑ two diseases within the next AI startup (since rebranded lion people worldwide. It has a two years. to Huma), has partnered wide range of symptoms and is with Chinese Internet com‑ therefore difficult to diagnose and pany, Tencent, to develop an monitor as it progresses. At-home diagnostics AI‑powered “at home” monitor‑ ing solution to assess the motor Medopad’s “at home” solution According to the World Health function of a Parkinson’s patient in monitors patients using wearable Organization, there will be a a real-world setting. devices and a smartphone app. global shortfall of 18 million Machine learning is then used to health workers by 2030. AI can “AI in health care is especially spot patterns in the data. help reduce this challenge by important as our healthcare helping doctors diagnose and system can’t cope with the “AI will take the burden off clini‑ assess patients with degener‑ increase in demand. We must cians and overstretched medical ative diseases such as cancer move towards a ’prevent not treat’ systems, while improving the lives and Parkinson’s Disease quickly model with predictive medicine of patients with Parkinson’s,” a and efficiently. at its core,” a spokesperson for Medopad spokesperson said. Medopad said in an interview. AI and health care ITU News MAGAZINE 02/2020 53
ITU and WHO “AI for Health” solutions “As we begin to accumulate larger pools of data, the possibili‑ ITU and WHO are developing systems for the performance ties to better understand disease benchmarking of “AI for Health” solutions. The initiative is open progression and risk become to all. Learn more here. endless. Mobile technologies are so ubiquitous that each individual is becoming a data centre for their own health.”
For patients, the technology means that tests can be done at home with a carer or family mem‑ ber. For doctors, it can reduce the time necessary to evaluate the patient, thereby improving efficiency and reducing costs.
ITU’s role One of the group’s key goals is to establish a standardized For doctors and healthcare assessment framework to eval‑ professionals, one of the next big uate AI‑based methods for challenges will be to adopt and health, diagnosis, triage or treat‑ integrate these new systems into ment decisions. daily clinical practice. ITU’s AI for Good Summit ITU’s Focus Group on AI for 2020 will look at how AI Health, a partnership with the can help to achieve the World Health Organization United Nations Sustainable (WHO), is empowering health‑ Development Goals. care workers with AI solutions to achieve universal healthcare coverage. AI and health care ITU News MAGAZINE 02/2020 54 Shutterstock
The opportunities and challenges of introducing AI to improve health care
By Chaesub Lee
Director of the ITU Telecommunication Standardization Bureau
J Participants at Kaleidoscope Technologies are converg‑ 2019 discussed the ing. Industries are converg‑ In AI for health, contribution of information and ing. Our world has become ITU and WHO have communication technologies highly interconnected. launched our most (ICTs) to health care, but this ambitious initiative yet. discussion is emblematic of a ITU standardization work is now larger trend. supporting digital transformation in areas ranging from energy Chaesub Lee We are living through a time of and transportation to financial Director of the ITU accelerating convergence. services; agriculture; smart cities; Standardization Bureau and, of course, health care. AI and health care ITU News MAGAZINE 02/2020 55
It is in this context that the ITU But in AI for health, ITU and WHO Like all software, AI solutions standardization platform, for have launched our most ambi‑ have a very dynamic update and many years central to building tious initiative yet. release cycle. Traditional certifica‑ trust within the ICT sector, is now tion processes will not scale. As a helping the ICT sector build trust result, AI models are not being with its many new partners — from AI for health certified and not being used. automotive to fintech to health care, and many more. Our Focus Group on AI for Health The benchmarking framework aims to establish the foundations being developed by the Focus Working in collaboration, think‑ for AI to improve health care on a Group will support the evolution ing in a multidisciplinary way, global scale. The Focus Group is of these certification processes. we can create very meaningful open to all. improvements in the quality of The group is studying AI use life enjoyed by billions of peo‑ We are working towards the cases for a range of health topics, ple worldwide. standardization of a framework including diagnostics chatbots, and associated processes for neuro-degenerative diseases, the benchmarking of “AI for histology, vision loss, skin Kaleidoscope 2019 Health” solutions. lesions, cardiovascular diseases, focused on ICT for health and psychiatry. We aim to create a facility open That is why ITU’s academic confer‑ to all innovators — an always-avail‑ And, of course, all these use cases ence is named “Kaleidoscope”. able online system for perfor‑ demand data. mance benchmarking. It views ICT innovation through We are bringing different indus‑ a kaleidoscope to see its many We see potential to deliver met‑ tries together. These industries dimensions. In the recent 2019 rics of great value to regulators. are governed by different reg‑ conference, we focused on ICT for ulations. And we have different health, but we approached our National health regulators are priorities and constraints when it discussions from a wide variety accustomed to testing diagnos‑ comes to data. of perspectives. tic tools such as microscopes. However, these regulators are It should come as no surprise, ITU standardization work for entering unfamiliar territory with then, that many see data govern‑ digital health receives essential the testing of methods such as ance as an issue of critical impor‑ support from our longstanding AI models. tance to the success of AI. partnership with the World Health Organization (WHO). In the AI era, data is the new commodity. AI and health care ITU News MAGAZINE 02/2020 56
Alongside discussions about Data: A driving force data privacy, we need to reflect for innovation In health, we frequently on what kind of commodity data hear, ’Do no harm’. should be: Data has become a driving force We are introducing for innovation across all industries AI into a field with Should it be like drinking and public-sector bodies. And no margin for error. water — abundant and free much like a natural resource, we for all? will draw the most value from data if we reach international Chaesub Lee Or like gold — relatively agreement on how to refine and Director of the ITU abundant but a share it. Standardization Bureau valuable resource? Kaleidoscope 2019, hosted by the Or like diamonds — where Georgia Institute of Technology supply is carefully controlled in Atlanta, United States from to maximize value? 4–6 December, with the theme: “ICT for health: Networks, In health, we frequently hear, Collective interests may standards and innovation,” “Do no harm”. We are introduc‑ prevail for public health, … made a valuable contribution to ing AI into a field with no margin but what about neglected and this discussion. for error. rare diseases? AI and health are natural com‑ This will be a true test of AI’s And how about, My data panions. We see huge oppor‑ maturity and a true test of AI’s versus Your data mindsets? tunities for AI to improve health potential to better our lives. care, but there are still significant An international dialogue challenges to be overcome. My Our experience introducing is needed to define rules remarks have focused on just two AI into the healthcare sector is of engagement for data, a of these challenges: benchmark‑ certain to inform AI innovation in “Commons” for data, to ensure ing and data governance. many other sectors. that all of humanity benefits from AI’s contribution to health care.
This is an area where a collabora‑ tive platform offered by ITU and WHO could deliver considera‑ ble value. AI and health care ITU News MAGAZINE 02/2020 57 Shutterstock
Responsibly unlocking the potential of data in health
By Stefan Germann and Ulla Jasper
Stefan Germann, CEO and Ulla Jasper, Policy Officer, Fondation Botnar
J We are witnessing a global This presents a great opportunity shift from traditional, reactive Young people have the to continue unleashing the poten‑ health care to proactive health potential to be powerful tial of AI and frontier technologies and wellbeing/wellness fuelled by changemakers, but for to improve health care. digital solutions. this to be realized they need to be included The uptake of digital solutions is A recently published article in the discussions welcome; however, it also raises from mHealth shows that 75% of about our collective critical ethical and societal ques‑ health-care companies are plan‑ digital future. tions that need answering. ning to execute artificial intelli‑ gence (AI) strategies in 2020. As technology becomes more Stefan Germann/ integrated in health services, Ulla Jasper there is and will continue to be, a CEO/Policy Officer, huge increase in both the genera‑ Fondation Botnar tion and usage of health data. AI and health care ITU News MAGAZINE 02/2020 58
Fondation Botnar — The future now, The current situation poses a risk for children worldwide for citizens’ privacy and we need to start developing governance structures that both aid innovation and protect our privacy.
Ownership of health data
Ownership of health data must be tackled before equity can be secured.
From a human rights perspective, ownership of health data can be problematic. There are many questions surrounding visibility and lack of privacy of citizens’ The opportunities to harness the “Our health data is constantly information, especially as the data transformative power of big data being recorded, stored, and have never been greater, but we shared, but the rules of how tech‑ We would like to see a different need to urgently come up with nology companies use citizens’ development where companies equitable and sustainable solu‑ sensitive data are few.” and government responsibly tions for health data governance. and transparently use citizens’ This, in combination with the data. However, while new power private sector taking an increas‑ structures over individuals, but Opportunities ingly leading role in designing also societies are emerging, there and challenges solutions for digital welfare, is a general reluctance among creates opportunities as well most governments to regulate Telemedicine, health chatbots/ as challenges. the activities of the big tech apps, and smart watches coupled companies. This is compounded with monitoring of social media There is a growing critical dia‑ by resistance from companies in and web data is bringing an logue around governance struc‑ taking systematic responsibility opportunity to leverage data to tures for health data, but it is time for human rights to be consid‑ better understand and generate for the dialogue to be turned into ered; new power structures over insights about health. concrete action. individuals are emerging, which has to be well-managed. AI and health care ITU News MAGAZINE 02/2020 59
More voice for to young people — many of whom At Fondation Botnar, young people are naturally digital natives — to we are prioritizing be part of shaping a fair and equi‑ the development With ongoing global efforts table future. and implementation on data and digital health pol‑ of digital solutions, icy initiatives, such as the UN Young people have the potential particularly in AI, Secretary’s High-Level Panel to be powerful changemakers, that improve the on Digital Collaboration, the but for this to be realized they health and wellbeing Financial Times and Lancet need to be included in the of young people in Commission ’Governing Health discussions about our collective low- and middle- Futures 2030: Growing up in a digital future. income countries. Digital World’, and the World Health Organization’s draft Global At Fondation Botnar, we are Strategy on Digital Health, we call prioritizing the development Stefan Germann/ on the global health community and implementation of digital Ulla Jasper to give more voice and attention solutions, particularly in AI, that CEO/Policy Officer, improve the health and wellbe‑ Fondation Botnar ing of young people in low- and middle-income countries.
We want to ensure young people are not just involved but have real tangible power and influence in the design of innovative solu‑ tions, and in shaping policy and global agendas.
These are some of the chal‑ lenges and opportunities we are currently facing. Join us at our session at the AI for Good Global Summit, where we will dive deeper into these chal‑ lenges and discuss opportunities together with young experts and rights specialists from around the world. Shutterstock Stay current // // Stay informed Sign up to: Key ICT trends worldwide Insights from ICT thought leaders The latest on ITU events and initiatives
Every Tuesday Regular blogs Six editions per year
Join ITU’s online communities on your favorite channel
Follow podcasts Receive latest news